The development of power MOSFET's was at least in part motivated by the objective of reducing the control current required by power bipolar devices during forced turn-off. In bipolar devices the injection of minority carriers into their drift region reduces the resistance to forward current flow. These devices are capable of operation at appreciable current densities, but are relatively inefficient as a consequence of the large currents required during device turn-on and turn-off.
In contrast, the gate structure of the power MOSFET has a very high steady-state impedance. This allows control of the device by a voltage source, since only relatively small gate drive currents are required to charge and discharge the input gate capacitance. Unfortunately, the ease of gating the power MOSFET is offset by its high on-state resistance arising from the absence of minority carrier injection. Hence, a combination of low-resistance bipolar-type current conduction with MOS gate control would provide the desired features of high operating forward current density and low gate drive power.
Referring to the cross-sectional illustration of FIG. 1, a device known as an insulated gate bipolar transistor (IGBT) illustrates one approach to combining these features. In this type of structure most of the forward current flow occurs between the emitter and collector terminals of the vertical PNP bipolar transistor portion of the device. The on-state losses of the IGBT at high voltages are significantly less than those of power MOSFET's due to the injection of minority carriers (electrons) into the N-base drift region.
As shown in FIG. 2, a regenerative device known as MOS-controlled thyristor (MCT) exhibits less forward voltage drop than does the IGBT. This P-N-P-N structure can be regarded as two transistors--an upper NPN transistor and a lower PNP transistor--that are internally connected in such a fashion as to obtain regenerative feedback between each other. Specifically, a thyristor may be considered as a combination of PNP and NPN bipolar transistors connected such that the base of each is driven by the collector current of the other. Once the thyristor is turned on via the gate electrode such that the requisite transistor turn-on current is supplied each transistor then drives the other into saturation. At this juncture the thyristor is no longer under the control of its gate electrode and continues to operate even in the absence of gate drive current--a phenomenon known as regenerative latch-up. Although the large current flow characteristic of regenerative on-state operation may be extinguished by appropriate modification of the gate voltage, the current produced by the MCT may not be adjusted (i.e., "fine-tuned") during regenerative operation through modulation of the gate voltage.
Since thyristors are often used in high-power switching applications, the maximum turn-off current level is generally of considerable importance. The MCT device of FIG. 2 is turned off by reversing the polarity of the applied gate voltage so as to eliminate the accumulation layer at the surface of the N-region embedded between the P and P+ regions underlying the gate. In this way a p-channel field-effect transistor (FET) within the device forms an active short circuit between the N+ cathode and P-base regions. The device will cease regenerative operation when the short-circuit current increases to the extent that the voltage across the N+/P junction falls below 0.7 V. Unfortunately, the maximum current which can be switched off by the MCT markedly decreases with increasing anode voltages at elevated temperatures. As a consequence, the current handling capability of the MCT has proven to be inadequate for particular circuit applications.
FIG. 3 depicts a four-layer semiconductor structure, generally termed a MOS gated emitter switched thyristor (EST), also designed to operate in a regenerative mode. When the gate voltage is at the cathode potential the device is in a forward blocking mode with the anode voltage supported across junction J1. The device is turned on by applying a positive bias to the gate to create a channel at the surface of the P-base region. As shown in FIG. 3, the regenerative thyristor portion of the device latches up when the junction between the N+ floating emitter and the P-base becomes forward biased. Regenerative on-state operation is extinguished by reducing the gate bias to zero, effectively disconnecting the emitter from the cathode. Again, however, current flow during on-state operation can be fine-tuned by modulation of the voltage applied to the gate electrode only until the breakdown voltage of the lateral MOSFET is reached. Because the channels of the MOS devices inherent within conventional EST structures are relatively short, the breakdown voltages thereof are typically less than 20 volts.
In addition, the N+/P junction of the conventional EST shown in FIG. 3 does not become reverse biased until the regenerative action of the main thyristor is sufficiently attenuated. It follows that the lateral MOSFET at the surface of the structure is prone to break down during high-voltage device deactivation as a consequence of supporting the large junction voltage. Moreover, the elevated hole current through the cathode which arises during turn-off of the device may induce undesired regenerative operation within a parasitic thyristor (see FIG. 3).
Accordingly, a need in the art exists for an emitter switched thyristor which:
(i) is disposed to be turned off rapidly (i.e., in less than 1 microsecond) without accompanying parasitic latch-up, and which PA1 (ii) allows the current provided thereby to be fine-tuned under gate control during regenerative on-state operation.